1548498 - artifact builds frequently failing on Try

Status: RESOLVED DUPLICATE of bug 1545387

(Developer Services :: General, defect)

Description

[Adrian Pop]

Can you please check the following links :

https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=244110638&repo=try&lineNumber=1109

and

https://treeherder.mozilla.org/#/jobs?repo=try&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel%2Crunnable&group_state=expanded&revision=cc4a2681fec82594e0081a0c2c84a04f0987f81a

and see if these are related to the changes to 40 digit shas in the revision on hg.mozilla.org ?

The artifact builds on Try are intermittently failing.

Comment 1

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

From:

https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=244110638&repo=try&lineNumber=1085

Try push is https://treeherder.mozilla.org/#/jobs?repo=try&revision=d77ab36d6a0357afb39e314c7016e8ee2735a56d&selectedJob=244110638
Parent is: https://hg.mozilla.org/try/rev/ee0dd3b092d03d28da65f5ee686db942b9ce8ec0
On central: https://hg.mozilla.org/mozilla-central/rev/ee0dd3b092d03d28da65f5ee686db942b9ce8ec0
Is that a short vs long revision issue or is anything else broken?

Comment 2

[Connor Sheehan [:sheehan]]

As far as I can tell, the 40-digit SHA change is unrelated to this bustage. Everywhere we query the log in-tree uses {node} as the revset (for example, here and here), meaning we were already using the 40-char SHA. On top of that, the pushlog JSON api has always served the 40-character SHA by default.

So I don't think the 40-char SHA caused this. If the number of failures is high I can look deeper, but I'll hold off on that until I have some numbers to justify the time spent.

Comment 3

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

For some pushes / at some times it's frequent, e.g. https://treeherder.mozilla.org/#/jobs?repo=try&resultStatus=pending%2Crunning%2Csuccess%2Ctestfailed%2Cbusted%2Cexception&searchStr=%28ba%29&tochange=ac1c81bedd856393597fafe8cb9f08cb7fd2e256&fromchange=a1b00dece83e1dad56f28cdadeae9364daaf6890 (also occurred yesterday around that time).

Here one push got 2 failures and earlier and later ones not: https://treeherder.mozilla.org/#/jobs?repo=try&resultStatus=pending%2Crunning%2Csuccess%2Ctestfailed%2Cbusted%2Cexception&searchStr=%28ba%29&tochange=f5dd91e283f9372b7ebd250e7eb79893438eca00&fromchange=604dfa7f1ec56888ca33aee32e557a8f023f1d31&selectedJob=244505432

Those failures cause the people to push again or request full builds.

Comment 4

[Kim Moir [:kmoir] ET]

Connor should this bug move to another bucket or should it remain here?

Comment 5

[Connor Sheehan [:sheehan]]

This is a duplicate of bug 1545387, which has been resolved. See bug 1545387 comment 7 for more info on what happened but tl;dr: the try repo on the private hg service didn't have the hgrc file replicated, causing revisions to be pulled down as public which made the revset of candidate revisions for an artifact build return the wrong changesets.